Merge pull request #2225 from dalf/processors

Processors

7 files changed, 701 insertions, 0 deletions

diff --git a/searx/search/__init__.py b/searx/search/__init__.py
new file mode 100644
index 000000000..4511463f3
--- /dev/null
+++ b/searx/search/__init__.py
@@ -0,0 +1,265 @@
+'''
+searx is free software: you can redistribute it and/or modify
+it under the terms of the GNU Affero General Public License as published by
+the Free Software Foundation, either version 3 of the License, or
+(at your option) any later version.
+
+searx is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+GNU Affero General Public License for more details.
+
+You should have received a copy of the GNU Affero General Public License
+along with searx. If not, see < http://www.gnu.org/licenses/ >.
+
+(C) 2013- by Adam Tauber, <asciimoo@gmail.com>
+'''
+
+import typing
+import gc
+import threading
+from time import time
+from uuid import uuid4
+from _thread import start_new_thread
+
+from searx import settings
+from searx.answerers import ask
+from searx.external_bang import get_bang_url
+from searx.results import ResultContainer
+from searx import logger
+from searx.plugins import plugins
+from searx.search.processors import processors, initialize as initialize_processors
+
+
+logger = logger.getChild('search')
+
+max_request_timeout = settings.get('outgoing', {}).get('max_request_timeout' or None)
+if max_request_timeout is None:
+    logger.info('max_request_timeout={0}'.format(max_request_timeout))
+else:
+    if isinstance(max_request_timeout, float):
+        logger.info('max_request_timeout={0} second(s)'.format(max_request_timeout))
+    else:
+        logger.critical('outgoing.max_request_timeout if defined has to be float')
+        import sys
+        sys.exit(1)
+
+
+def initialize(settings_engines=None):
+    settings_engines = settings_engines or settings['engines']
+    initialize_processors(settings_engines)
+
+
+class EngineRef:
+
+    __slots__ = 'name', 'category', 'from_bang'
+
+    def __init__(self, name: str, category: str, from_bang: bool=False):
+        self.name = name
+        self.category = category
+        self.from_bang = from_bang
+
+    def __repr__(self):
+        return "EngineRef({!r}, {!r}, {!r})".format(self.name, self.category, self.from_bang)
+
+    def __eq__(self, other):
+        return self.name == other.name and self.category == other.category and self.from_bang == other.from_bang
+
+
+class SearchQuery:
+    """container for all the search parameters (query, language, etc...)"""
+
+    __slots__ = 'query', 'engineref_list', 'categories', 'lang', 'safesearch', 'pageno', 'time_range',\
+                'timeout_limit', 'external_bang'
+
+    def __init__(self,
+                 query: str,
+                 engineref_list: typing.List[EngineRef],
+                 categories: typing.List[str],
+                 lang: str,
+                 safesearch: int,
+                 pageno: int,
+                 time_range: typing.Optional[str],
+                 timeout_limit: typing.Optional[float]=None,
+                 external_bang: typing.Optional[str]=None):
+        self.query = query
+        self.engineref_list = engineref_list
+        self.categories = categories
+        self.lang = lang
+        self.safesearch = safesearch
+        self.pageno = pageno
+        self.time_range = time_range
+        self.timeout_limit = timeout_limit
+        self.external_bang = external_bang
+
+    def __repr__(self):
+        return "SearchQuery({!r}, {!r}, {!r}, {!r}, {!r}, {!r}, {!r}, {!r}, {!r})".\
+               format(self.query, self.engineref_list, self.categories, self.lang, self.safesearch,
+                      self.pageno, self.time_range, self.timeout_limit, self.external_bang)
+
+    def __eq__(self, other):
+        return self.query == other.query\
+            and self.engineref_list == other.engineref_list\
+            and self.categories == self.categories\
+            and self.lang == other.lang\
+            and self.safesearch == other.safesearch\
+            and self.pageno == other.pageno\
+            and self.time_range == other.time_range\
+            and self.timeout_limit == other.timeout_limit\
+            and self.external_bang == other.external_bang
+
+
+class Search:
+    """Search information container"""
+
+    __slots__ = "search_query", "result_container", "start_time", "actual_timeout"
+
+    def __init__(self, search_query):
+        # init vars
+        super().__init__()
+        self.search_query = search_query
+        self.result_container = ResultContainer()
+        self.start_time = None
+        self.actual_timeout = None
+
+    def search_external_bang(self):
+        """
+        Check if there is a external bang.
+        If yes, update self.result_container and return True
+        """
+        if self.search_query.external_bang:
+            self.result_container.redirect_url = get_bang_url(self.search_query)
+
+            # This means there was a valid bang and the
+            # rest of the search does not need to be continued
+            if isinstance(self.result_container.redirect_url, str):
+                return True
+        return False
+
+    def search_answerers(self):
+        """
+        Check if an answer return a result.
+        If yes, update self.result_container and return True
+        """
+        answerers_results = ask(self.search_query)
+
+        if answerers_results:
+            for results in answerers_results:
+                self.result_container.extend('answer', results)
+            return True
+        return False
+
+    # do search-request
+    def _get_requests(self):
+        # init vars
+        requests = []
+
+        # max of all selected engine timeout
+        default_timeout = 0
+
+        # start search-reqest for all selected engines
+        for engineref in self.search_query.engineref_list:
+            processor = processors[engineref.name]
+
+            # set default request parameters
+            request_params = processor.get_params(self.search_query, engineref.category)
+            if request_params is None:
+                continue
+
+            # append request to list
+            requests.append((engineref.name, self.search_query.query, request_params))
+
+            # update default_timeout
+            default_timeout = max(default_timeout, processor.engine.timeout)
+
+        # adjust timeout
+        actual_timeout = default_timeout
+        query_timeout = self.search_query.timeout_limit
+
+        if max_request_timeout is None and query_timeout is None:
+            # No max, no user query: default_timeout
+            pass
+        elif max_request_timeout is None and query_timeout is not None:
+            # No max, but user query: From user query except if above default
+            actual_timeout = min(default_timeout, query_timeout)
+        elif max_request_timeout is not None and query_timeout is None:
+            # Max, no user query: Default except if above max
+            actual_timeout = min(default_timeout, max_request_timeout)
+        elif max_request_timeout is not None and query_timeout is not None:
+            # Max & user query: From user query except if above max
+            actual_timeout = min(query_timeout, max_request_timeout)
+
+        logger.debug("actual_timeout={0} (default_timeout={1}, ?timeout_limit={2}, max_request_timeout={3})"
+                     .format(actual_timeout, default_timeout, query_timeout, max_request_timeout))
+
+        return requests, actual_timeout
+
+    def search_multiple_requests(self, requests):
+        search_id = uuid4().__str__()
+
+        for engine_name, query, request_params in requests:
+            th = threading.Thread(
+                target=processors[engine_name].search,
+                args=(query, request_params, self.result_container, self.start_time, self.actual_timeout),
+                name=search_id,
+            )
+            th._timeout = False
+            th._engine_name = engine_name
+            th.start()
+
+        for th in threading.enumerate():
+            if th.name == search_id:
+                remaining_time = max(0.0, self.actual_timeout - (time() - self.start_time))
+                th.join(remaining_time)
+                if th.is_alive():
+                    th._timeout = True
+                    self.result_container.add_unresponsive_engine(th._engine_name, 'timeout')
+                    logger.warning('engine timeout: {0}'.format(th._engine_name))
+
+    def search_standard(self):
+        """
+        Update self.result_container, self.actual_timeout
+        """
+        requests, self.actual_timeout = self._get_requests()
+
+        # send all search-request
+        if requests:
+            self.search_multiple_requests(requests)
+            start_new_thread(gc.collect, tuple())
+
+        # return results, suggestions, answers and infoboxes
+        return True
+
+    # do search-request
+    def search(self):
+        self.start_time = time()
+
+        if not self.search_external_bang():
+            if not self.search_answerers():
+                self.search_standard()
+
+        return self.result_container
+
+
+class SearchWithPlugins(Search):
+    """Similar to the Search class but call the plugins."""
+
+    __slots__ = 'ordered_plugin_list', 'request'
+
+    def __init__(self, search_query, ordered_plugin_list, request):
+        super().__init__(search_query)
+        self.ordered_plugin_list = ordered_plugin_list
+        self.request = request
+
+    def search(self):
+        if plugins.call(self.ordered_plugin_list, 'pre_search', self.request, self):
+            super().search()
+
+        plugins.call(self.ordered_plugin_list, 'post_search', self.request, self)
+
+        results = self.result_container.get_ordered_results()
+
+        for result in results:
+            plugins.call(self.ordered_plugin_list, 'on_result', self.request, self, result)
+
+        return self.result_container
diff --git a/searx/search/processors/__init__.py b/searx/search/processors/__init__.py
new file mode 100644
index 000000000..4cae3cd0f
--- /dev/null
+++ b/searx/search/processors/__init__.py
@@ -0,0 +1,41 @@
+# SPDX-License-Identifier: AGPL-3.0-or-later
+
+from .online import OnlineProcessor
+from .offline import OfflineProcessor
+from .online_dictionary import OnlineDictionaryProcessor
+from .online_currency import OnlineCurrencyProcessor
+from .abstract import EngineProcessor
+from searx import logger
+import searx.engines as engines
+
+
+__all__ = ['EngineProcessor', 'OfflineProcessor', 'OnlineProcessor',
+           'OnlineDictionaryProcessor', 'OnlineCurrencyProcessor', 'processors']
+logger = logger.getChild('search.processors')
+processors = {}
+
+
+def get_processor_class(engine_type):
+    for c in [OnlineProcessor, OfflineProcessor, OnlineDictionaryProcessor, OnlineCurrencyProcessor]:
+        if c.engine_type == engine_type:
+            return c
+    return None
+
+
+def get_processor(engine, engine_name):
+    engine_type = getattr(engine, 'engine_type', 'online')
+    processor_class = get_processor_class(engine_type)
+    if processor_class:
+        return processor_class(engine, engine_name)
+    else:
+        return None
+
+
+def initialize(engine_list):
+    engines.initialize_engines(engine_list)
+    for engine_name, engine in engines.engines.items():
+        processor = get_processor(engine, engine_name)
+        if processor is None:
+            logger.error('Error get processor for engine %s', engine_name)
+        else:
+            processors[engine_name] = processor
diff --git a/searx/search/processors/abstract.py b/searx/search/processors/abstract.py
new file mode 100644
index 000000000..cf3fd7236
--- /dev/null
+++ b/searx/search/processors/abstract.py
@@ -0,0 +1,39 @@
+# SPDX-License-Identifier: AGPL-3.0-or-later
+
+from abc import abstractmethod
+from searx import logger
+
+
+logger = logger.getChild('searx.search.processor')
+
+
+class EngineProcessor:
+
+    def __init__(self, engine, engine_name):
+        self.engine = engine
+        self.engine_name = engine_name
+
+    def get_params(self, search_query, engine_category):
+        # if paging is not supported, skip
+        if search_query.pageno > 1 and not self.engine.paging:
+            return None
+
+        # if time_range is not supported, skip
+        if search_query.time_range and not self.engine.time_range_support:
+            return None
+
+        params = {}
+        params['category'] = engine_category
+        params['pageno'] = search_query.pageno
+        params['safesearch'] = search_query.safesearch
+        params['time_range'] = search_query.time_range
+
+        if hasattr(self.engine, 'language') and self.engine.language:
+            params['language'] = self.engine.language
+        else:
+            params['language'] = search_query.lang
+        return params
+
+    @abstractmethod
+    def search(self, query, params, result_container, start_time, timeout_limit):
+        pass
diff --git a/searx/search/processors/offline.py b/searx/search/processors/offline.py
new file mode 100644
index 000000000..ede8eb5e1
--- /dev/null
+++ b/searx/search/processors/offline.py
@@ -0,0 +1,51 @@
+# SPDX-License-Identifier: AGPL-3.0-or-later
+
+import threading
+from time import time
+from searx import logger
+from searx.metrology.error_recorder import record_exception, record_error
+from searx.search.processors.abstract import EngineProcessor
+
+
+logger = logger.getChild('search.processor.offline')
+
+
+class OfflineProcessor(EngineProcessor):
+
+    engine_type = 'offline'
+
+    def _record_stats_on_error(self, result_container, start_time):
+        engine_time = time() - start_time
+        result_container.add_timing(self.engine_name, engine_time, engine_time)
+
+        with threading.RLock():
+            self.engine.stats['errors'] += 1
+
+    def _search_basic(self, query, params):
+        return self.engine.search(query, params)
+
+    def search(self, query, params, result_container, start_time, timeout_limit):
+        try:
+            search_results = self._search_basic(query, params)
+
+            if search_results:
+                result_container.extend(self.engine_name, search_results)
+
+                engine_time = time() - start_time
+                result_container.add_timing(self.engine_name, engine_time, engine_time)
+                with threading.RLock():
+                    self.engine.stats['engine_time'] += engine_time
+                    self.engine.stats['engine_time_count'] += 1
+
+        except ValueError as e:
+            record_exception(self.engine_name, e)
+            self._record_stats_on_error(result_container, start_time)
+            logger.exception('engine {0} : invalid input : {1}'.format(self.engine_name, e))
+        except Exception as e:
+            record_exception(self.engine_name, e)
+            self._record_stats_on_error(result_container, start_time)
+            result_container.add_unresponsive_engine(self.engine_name, 'unexpected crash', str(e))
+            logger.exception('engine {0} : exception : {1}'.format(self.engine_name, e))
+        else:
+            if getattr(threading.current_thread(), '_timeout', False):
+                record_error(self.engine_name, 'Timeout')
diff --git a/searx/search/processors/online.py b/searx/search/processors/online.py
new file mode 100644
index 000000000..17bbfef0f
--- /dev/null
+++ b/searx/search/processors/online.py
@@ -0,0 +1,211 @@
+# SPDX-License-Identifier: AGPL-3.0-or-later
+
+from urllib.parse import urlparse
+from time import time
+import threading
+
+import requests.exceptions
+
+import searx.poolrequests as poolrequests
+from searx.engines import settings
+from searx import logger
+from searx.utils import gen_useragent
+from searx.exceptions import (SearxEngineAccessDeniedException, SearxEngineCaptchaException,
+                              SearxEngineTooManyRequestsException,)
+from searx.metrology.error_recorder import record_exception, record_error
+
+from searx.search.processors.abstract import EngineProcessor
+
+
+logger = logger.getChild('search.processor.online')
+
+DEFAULT_PARAMS = {
+    'method': 'GET',
+    'headers': {},
+    'data': {},
+    'url': '',
+    'cookies': {},
+    'verify': True,
+    'auth': None
+}
+
+
+class OnlineProcessor(EngineProcessor):
+
+    engine_type = 'online'
+
+    def get_params(self, search_query, engine_category):
+        params = super().get_params(search_query, engine_category)
+        if params is None:
+            return None
+
+        # skip suspended engines
+        if self.engine.suspend_end_time >= time():
+            logger.debug('Engine currently suspended: %s', self.engine_name)
+            return None
+
+        # add default params
+        params.update(DEFAULT_PARAMS)
+
+        # add an user agent
+        params['headers']['User-Agent'] = gen_useragent()
+
+        return params
+
+    def _send_http_request(self, params):
+        # create dictionary which contain all
+        # informations about the request
+        request_args = dict(
+            headers=params['headers'],
+            cookies=params['cookies'],
+            verify=params['verify'],
+            auth=params['auth']
+        )
+
+        # setting engine based proxies
+        if hasattr(self.engine, 'proxies'):
+            request_args['proxies'] = poolrequests.get_proxies(self.engine.proxies)
+
+        # max_redirects
+        max_redirects = params.get('max_redirects')
+        if max_redirects:
+            request_args['max_redirects'] = max_redirects
+
+        # soft_max_redirects
+        soft_max_redirects = params.get('soft_max_redirects', max_redirects or 0)
+
+        # raise_for_status
+        request_args['raise_for_httperror'] = params.get('raise_for_httperror', False)
+
+        # specific type of request (GET or POST)
+        if params['method'] == 'GET':
+            req = poolrequests.get
+        else:
+            req = poolrequests.post
+
+        request_args['data'] = params['data']
+
+        # send the request
+        response = req(params['url'], **request_args)
+
+        # check soft limit of the redirect count
+        if len(response.history) > soft_max_redirects:
+            # unexpected redirect : record an error
+            # but the engine might still return valid results.
+            status_code = str(response.status_code or '')
+            reason = response.reason or ''
+            hostname = str(urlparse(response.url or '').netloc)
+            record_error(self.engine_name,
+                         '{} redirects, maximum: {}'.format(len(response.history), soft_max_redirects),
+                         (status_code, reason, hostname))
+
+        return response
+
+    def _search_basic(self, query, params):
+        # update request parameters dependent on
+        # search-engine (contained in engines folder)
+        self.engine.request(query, params)
+
+        # ignoring empty urls
+        if params['url'] is None:
+            return None
+
+        if not params['url']:
+            return None
+
+        # send request
+        response = self._send_http_request(params)
+
+        # parse the response
+        response.search_params = params
+        return self.engine.response(response)
+
+    def search(self, query, params, result_container, start_time, timeout_limit):
+        # set timeout for all HTTP requests
+        poolrequests.set_timeout_for_thread(timeout_limit, start_time=start_time)
+        # reset the HTTP total time
+        poolrequests.reset_time_for_thread()
+
+        # suppose everything will be alright
+        requests_exception = False
+        suspended_time = None
+
+        try:
+            # send requests and parse the results
+            search_results = self._search_basic(query, params)
+
+            # check if the engine accepted the request
+            if search_results is not None:
+                # yes, so add results
+                result_container.extend(self.engine_name, search_results)
+
+                # update engine time when there is no exception
+                engine_time = time() - start_time
+                page_load_time = poolrequests.get_time_for_thread()
+                result_container.add_timing(self.engine_name, engine_time, page_load_time)
+                with threading.RLock():
+                    self.engine.stats['engine_time'] += engine_time
+                    self.engine.stats['engine_time_count'] += 1
+                    # update stats with the total HTTP time
+                    self.engine.stats['page_load_time'] += page_load_time
+                    self.engine.stats['page_load_count'] += 1
+        except Exception as e:
+            record_exception(self.engine_name, e)
+
+            # Timing
+            engine_time = time() - start_time
+            page_load_time = poolrequests.get_time_for_thread()
+            result_container.add_timing(self.engine_name, engine_time, page_load_time)
+
+            # Record the errors
+            with threading.RLock():
+                self.engine.stats['errors'] += 1
+
+            if (issubclass(e.__class__, requests.exceptions.Timeout)):
+                result_container.add_unresponsive_engine(self.engine_name, 'HTTP timeout')
+                # requests timeout (connect or read)
+                logger.error("engine {0} : HTTP requests timeout"
+                             "(search duration : {1} s, timeout: {2} s) : {3}"
+                             .format(self.engine_name, engine_time, timeout_limit, e.__class__.__name__))
+                requests_exception = True
+            elif (issubclass(e.__class__, requests.exceptions.RequestException)):
+                result_container.add_unresponsive_engine(self.engine_name, 'HTTP error')
+                # other requests exception
+                logger.exception("engine {0} : requests exception"
+                                 "(search duration : {1} s, timeout: {2} s) : {3}"
+                                 .format(self.engine_name, engine_time, timeout_limit, e))
+                requests_exception = True
+            elif (issubclass(e.__class__, SearxEngineCaptchaException)):
+                result_container.add_unresponsive_engine(self.engine_name, 'CAPTCHA required')
+                logger.exception('engine {0} : CAPTCHA')
+                suspended_time = e.suspended_time  # pylint: disable=no-member
+            elif (issubclass(e.__class__, SearxEngineTooManyRequestsException)):
+                result_container.add_unresponsive_engine(self.engine_name, 'too many requests')
+                logger.exception('engine {0} : Too many requests')
+                suspended_time = e.suspended_time  # pylint: disable=no-member
+            elif (issubclass(e.__class__, SearxEngineAccessDeniedException)):
+                result_container.add_unresponsive_engine(self.engine_name, 'blocked')
+                logger.exception('engine {0} : Searx is blocked')
+                suspended_time = e.suspended_time  # pylint: disable=no-member
+            else:
+                result_container.add_unresponsive_engine(self.engine_name, 'unexpected crash')
+                # others errors
+                logger.exception('engine {0} : exception : {1}'.format(self.engine_name, e))
+        else:
+            if getattr(threading.current_thread(), '_timeout', False):
+                record_error(self.engine_name, 'Timeout')
+
+        # suspend the engine if there is an HTTP error
+        # or suspended_time is defined
+        with threading.RLock():
+            if requests_exception or suspended_time:
+                # update continuous_errors / suspend_end_time
+                self.engine.continuous_errors += 1
+                if suspended_time is None:
+                    suspended_time = min(settings['search']['max_ban_time_on_fail'],
+                                         self.engine.continuous_errors * settings['search']['ban_time_on_fail'])
+                self.engine.suspend_end_time = time() + suspended_time
+            else:
+                # reset the suspend variables
+                self.engine.continuous_errors = 0
+                self.engine.suspend_end_time = 0
diff --git a/searx/search/processors/online_currency.py b/searx/search/processors/online_currency.py
new file mode 100644
index 000000000..f0e919c03
--- /dev/null
+++ b/searx/search/processors/online_currency.py
@@ -0,0 +1,57 @@
+# SPDX-License-Identifier: AGPL-3.0-or-later
+
+import unicodedata
+import re
+
+from searx.data import CURRENCIES
+from .online import OnlineProcessor
+
+
+parser_re = re.compile('.*?(\\d+(?:\\.\\d+)?) ([^.0-9]+) (?:in|to) ([^.0-9]+)', re.I)
+
+
+def normalize_name(name):
+    name = name.lower().replace('-', ' ').rstrip('s')
+    name = re.sub(' +', ' ', name)
+    return unicodedata.normalize('NFKD', name).lower()
+
+
+def name_to_iso4217(name):
+    global CURRENCIES
+    name = normalize_name(name)
+    currency = CURRENCIES['names'].get(name, [name])
+    return currency[0]
+
+
+def iso4217_to_name(iso4217, language):
+    global CURRENCIES
+    return CURRENCIES['iso4217'].get(iso4217, {}).get(language, iso4217)
+
+
+class OnlineCurrencyProcessor(OnlineProcessor):
+
+    engine_type = 'online_currency'
+
+    def get_params(self, search_query, engine_category):
+        params = super().get_params(search_query, engine_category)
+        if params is None:
+            return None
+
+        m = parser_re.match(search_query.query)
+        if not m:
+            return None
+
+        amount_str, from_currency, to_currency = m.groups()
+        try:
+            amount = float(amount_str)
+        except ValueError:
+            return None
+        from_currency = name_to_iso4217(from_currency.strip())
+        to_currency = name_to_iso4217(to_currency.strip())
+
+        params['amount'] = amount
+        params['from'] = from_currency
+        params['to'] = to_currency
+        params['from_name'] = iso4217_to_name(from_currency, 'en')
+        params['to_name'] = iso4217_to_name(to_currency, 'en')
+        return params
diff --git a/searx/search/processors/online_dictionary.py b/searx/search/processors/online_dictionary.py
new file mode 100644
index 000000000..8e9ef1620
--- /dev/null
+++ b/searx/search/processors/online_dictionary.py
@@ -0,0 +1,37 @@
+# SPDX-License-Identifier: AGPL-3.0-or-later
+
+import re
+
+from searx.utils import is_valid_lang
+from .online import OnlineProcessor
+
+
+parser_re = re.compile('.*?([a-z]+)-([a-z]+) ([^ ]+)$', re.I)
+
+
+class OnlineDictionaryProcessor(OnlineProcessor):
+
+    engine_type = 'online_dictionnary'
+
+    def get_params(self, search_query, engine_category):
+        params = super().get_params(search_query, engine_category)
+        if params is None:
+            return None
+
+        m = parser_re.match(search_query.query)
+        if not m:
+            return None
+
+        from_lang, to_lang, query = m.groups()
+
+        from_lang = is_valid_lang(from_lang)
+        to_lang = is_valid_lang(to_lang)
+
+        if not from_lang or not to_lang:
+            return None
+
+        params['from_lang'] = from_lang
+        params['to_lang'] = to_lang
+        params['query'] = query
+
+        return params